3.2 Methods and Materials
3.2.3 The Kanji-picture Mapping Task
As mentioned above, no previous research on kanji acquisition has been conducted with pictures as one of the tasks to test L2 learners’ kanji recognition. However, pictures can provide pictorial clues; therefore, a kanji-picture mapping task with multiple choices was employed to test the participants’ kanji knowledge. The participants in this task were asked to map pictures with correct one-kanji character in Part 1 and two-kanji characters in Part 2 (see Appendix A).
Table 3-4 A Test Example in Pictorial Format
No. Answer Pictures Choices
Part 1 (one-kanji character)
(A) (A) 足 (B) 踝 (C) 支
Part 2 (two-kanji character)
(A) (A) 大根 (B) 萝卜 (C)胡卜
As can be seen in Table 3-4, the task was designed in multiple choice format, with three options provided for each question. The participants had to choose one correct answer and write the answer in the column.
3.2.4 Kanji-text Mapping Task (TM Task)
The second task is a kanji-text mapping task. According to Mori (2003), context can offer participants more meaning-related information. In their task, textual clues were given and helped the participants judge the correct meaning of kanji. Hence, the task design of one-kanji and two-kanji characters in textual format is shown in Table 3-5.
Table 3-5 A Test Example in Textual Format
No. Answer Fill in the blank Choices
This task was also designed in the multiple choice format. There were three options for each question. The participants had to choose one correct answer to the test sentence in textual format and write the answer in the column (see Appendix B).
3.3 Procedures
The procedures of a pilot study and the formal study are reported in this section. The scoring and statistical analysis of the formal study is discussed as well.
3.3.1 The Pilot Study
Six participants graduated from the college were recruited in the pilot study. They were grouped into basic-level (2 participants), intermediate-level (2 participants), and advanced-level (2 participants) according to their performance on the Japanese Language Proficiency Test. In the kanji-picture mapping task, the participants were asked to map the pictures with correct one-kanji and two-kanji characters. They continued to complete the
kanji-text mapping task to judge which one-kanji and two-kanji characters could fill in the
sentences appropriately. Each correct answer to a question was given one point and a wrong one received zero point.The major findings of the pilot study are illustrated as follows (see Appendix C). In response to the form correspondence, the results of the three types of forms of kanji characters were that the scores for kanji characters in same forms and semi-same forms were higher than different forms, indicating that the participants did rely on the visual information, and same forms could help our L2 learners interpret kanji characters. This result supported Chikamatsu’s (1996) findings. Besides, the advanced participants performed the best while the basic and intermediate participants performed similarly.
With regard to the meaning correspondence, kanji characters with same meanings were scored significantly higher than kanji characters with different meanings. As Vee (2006) suggested, kanji characters are semantic in nature. Thus, the meaning correspondence was found influential to our participants. It was found that the mean scores of the experimental groups gradually increased. Hence, L2 proficiency positively correlated with the meaning of
kanji characters.
With respect to the types, the difficulty order was SsF-SM > SF-SM > SFDM = SsF-DM. In this pattern, it was found that our participants performed better when the types were with same meanings, showing that meanings were more dominant than forms. Once again, this result proved the importance of meanings. The hypotheses of Lado (1957) and
Stockwell et al. (1965) were also supported that the differences between the L1 and L2 were difficult for learners to acquire. An increase on mean scores of three proficiency groups was also found.
As for the tasks, the participants’ performances on the two tasks were close, but the TM task received higher scores, indicating that the pictorial cues were more useful for interpreting kanji characters for our Chinese learners. This result matched Haththotuwa Gamage (2003) that as one of the L2 learners’ learner strategy, they considered pictorial cues were more efficient. Besides, our participants’ performance on two tasks also varied with their L2 proficiency.
Nonetheless, there were two limitations to the pilot study. First, some items in the multiple choice format seemed ambiguous and the pictures presented in the task were not clear enough. Therefore, the test items need to be revised. Second, the pool of the participants was too small and the results of proficiency in the two tasks were not consistent.
Hence, in the formal study more participants need to be recruited.
3.3.2 The Formal Study
As there were some inadequacies in the pilot study, the test items were revised in the formal study. As mentioned above, seventy-two Chinese learners of Japanese at three different proficiency levels were asked to participate in the present study from Soochow University and Kainan University. There were also twenty-four Japanese native speakers as a baseline group. Two tasks were implemented. The procedure of the formal study illustrates in Figure 3-1.
The Consent Form Assignment
2 minutes
Task 1: Kanji-picture Mapping Task
15 minutes
Task 2: Kanji-text Mapping Task
15 minutes
Figure 3-1 The Procedure of the Present Study
Each participant had to sign a consent form (see Appendix D). This step lasted only 2 minutes. After that, they were asked to do the two tasks. The kanji-picture mapping task and the kanji-text mapping task took the participants thirty minutes to complete.
3.3.3 Scoring and Statistical Analysis
In the kanji-picture mapping task and the kanji-text mapping task, the participants were asked to choose one best answer which could denote the meaning presented in the pictures and could complete the sentence from the multiple choices, respectively. The accuracy rate was calculated. Each correct answer was given one point. If the answer was wrong, zero point was given. Besides the scoring, the software of SPSS was implemented to run statistical reliability and significance of the results. The overall mean scores and standard deviations were calculated from the statistical analysis of software. Moreover, one-way analysis of variance (ANOVA) was applied to examine the differences among the three groups and the native baseline group.
3.4 Summary of Chapter Three
This chapter introduced the research design of the present study, including the participants, method and materials, a new classification of kanji characters (regarding the relation between forms and meanings), the procedures and the scoring and statistic analysis.
The participants were 72 Japanese majors in Taiwan as the experimental group and 24
native adults as the baseline group. The experimental group was divided into three groups according to their performance on the standardized Japanese proficiency test. The present study conducted a kanji-picture mapping task and a kanji-textual mapping task, both of which were presented in two parts (i.e., one-kanji characters and two-kanji characters).
C HAPTER F OUR
R ESULTS AND D ISCUSSION
The results of the present study are presented and discussed in this chapter. Section 4.1 examines the results of the form effect of Japanese kanji characters, and Section 4.2 discusses the influence of the meaning effect on Chinese characters and kanji characters on Chinese learners of Japanese. Section 4.3 explores Chinese learners’ performance and the learning difficulty of four form-meaning-integrated types of Japanese kanji characters. Section 4.4 analyzes the task effects (pictorial cues and textual cues) of the study, and Section 4.5 reports the effects of L2 proficiency on the interpretation of Japanese kanji characters. Finally, this chapter is summarized in Section 4.6.
4.1 Form-driven Effects
The first research question discusses the form correspondence between Japanese kanji characters and Chinese characters. The overall performances of the participants on three types of forms and a general discussion about the form effect are as follows.
Table 4-1 Participants’ Overall Performances on the Three Types of Forms of Japanese
Kanji Characters
Type M SD F
p-value
same form 0.86 0.11 33.916b .000***
semi-same form 0.83 0.11
different form 0.73 0.21
In Table 4-1, the overall results of the participants’ performances are presented. According to the results of one-way ANOVA, it was found that there was a significant difference among the same form, the semi-same form and the different form of Japanese kanji characters (F (2, 91) = 33.916b, p < .001) from both tasks. The Scheffé post hoc analysis showed that our
participants performed significantly better on the same form than that on the semi-same form, and that their performance on semi-same form was better than the different form (same form:
M = 0.86 > semi-same form: M = 0.83 > different form: M = 0.73). It was found that our participants scored differently on the three types of forms.
Table 4-2 Participants’ Overall Performances on the Two Types of Forms (combining same and semi-same forms) of Japanese Kanji Characters
Type M SD F
p-value
same + semi-same form 0.85 0.10 57.579b .000***
different form 0.73 0.20
As shown in the Table 4-2, when we combined the results of the same form and the semi-same form and compared them as whole with the different form, the result was also significant. The mean score of the same form and the semi-same form together could reach as high as 0.85.
Table 4-3 further presents the within-group differences among three types of forms, and Table 4-4 shows the results of the Scheffé post hoc analysis.
Table 4-3 The p-values for the within-group Differences among Three Types of Forms of Japanese Kanji Characters
B I A N
F
14.824 6.319 6.564 6.917P
.000*** .003** .002** .002***Table 4-4 The Results of the Scheffé Post Hoc Analysis (Comparison between Three Types of
Std. Error Sig. 95% Confidence Interval Lower Bound Upper Bound
*. The mean difference is significant at the 0.05 level.
The Scheffé post hoc analysis showed that the same form (p < .001) and the semi-same form (p < .001) were significantly easier to acquire than the different form for Group B (F (2, 69)
= 14.824, p < .001). The same form (p < .01) and the semi-same form (p < .05) were also significantly easier to acquire than the different form for Group I (F (2, 69) = 6.319, p < .01).
As for Group A (F (2, 69) = 6.564, p < .01), the same form was significantly easier to acquire
than the different form (p < .01). However, there were no significant results between the semi-same form and the different form and between the same form and the semi-same form for Group A. Lastly, the same form (p < .01) was easier to acquire than the different form for the native baseline group (F (2, 69) = 6.917, p < .01). The participants’ performances on the same form and the different form and those on the same form and the semi-same form were not significantly different.
Therefore, the results of the participants’ overall performances and the within-group comparison showed that the difficulty order of the three types of form of kanji characters (from easy to difficult) was same form > semi-same form > different form. While the different form was the most difficult for all the participants, the forms with the same parts (i.e., the same form and the semi-same form) were easier to understand and interpret.
In the five-type hierarchy of difficulty (Stockwell et al., 1965, the simplified version adapted from Larsen-Freeman & Long, 1991, p.54), the Correspondence type was the easiest, whereas the Split was the most difficult, as can be seen in Table 4-5.
Table 4-5 Hierarchy of Difficulty (same as Table 1-2 on p.8) Degree Type of Difficulty
Easy
Hard
1. Correspondence 2. Coalesced 3. Absent 4. New 5. Split
The correspondence type refers to a function or a grammatical structure which exists in both L1 and L2 and this type conforms to the same form in the present study. It was also found that the same form was the easiest type to acquire. However, it was found that the five-type difficulty could not fully explain our examined types because they did not identify the semi-correspondence situation, that is, the semi-same form in the present study. The kanji
characters of the semi-same form are similar to the L1 form but not totally identical.
Therefore, a new type, i.e., Semi-correspondence type was added, as shown in Table 4-6. The Semi-correspondence occurs when the L1 and L2 are half-identical. There is no compatible type for our different meaning as well. The function exists in both languages, but in Japanese these kanji characters have different meanings or forms. Therefore, another new type was added, Retranslated, into the hierarchy of difficulty. The hierarchy of difficulty is revised from five types into seven types.
Table 4-6 Revised Hierarchy of Difficulty and the Examples from the Tasks Degree Type of Difficulty Examples from the Tasks
Easy
Hard
1. Correspondence 類 ‘type’: 獅子と虎は貓の類だ。
(TM Task, Part 1, Q1) 2. Semi-correspondence
換気‘to ventilate’: 窓をあけて換気をする。室 内の汚れた空気を、新鮮な空気と入れ替える こと。 (TM Task, Part 2, Q6)
3. Coalesced n/a
4. Absent n/a
5. Retranslated 波風 ‘argument’: 羽田の家は、波風の立たな
い円満な家庭。 (TM Task, Part 2, Q9)
6. New n/a
7. Split n/a
As can be seen in Table 4-6, our same form (e.g. 類 ‘type’) conforms to the Correspondence type and this type is the easiest. The form of kanji characters 類 is the same as the Chinese one that our learners of Japanese could find its corresponding form and transfer this same form into the L2. The semi-same form corresponds to the second Semi-correspondence type, since the semi-same features can still be identified and these identified features might be transferred when our L2 Japanese learners acquired kanji characters. In the example, the form of kanji characters 換気 is half similar to 換氣 in the Chinese characters and hence this
similarity may be transferred into L2. The third different form conforms to the Retranslated type in that an item in the L1 will be retranslated or reshaped in the L2. The form of Chinese characters of ‘argument’ is 風波, but it has been reformed as 波風.
Based on our findings, our revised hierarchy of difficulty was supported. If the same item in forms in the L1 and L2 could be identified as the Correspondence type, it could facilitate L2 acquisition. In Zimmermann’s (1987) study, he also found that the similarity in forms was usually expected to be perceived first. When the similarity is verified by the L2 learners, they assume that there would be a similarity in word meanings and would make more correct responses. Besides the same form, the mean score of the semi-same form was closer to that of the same form rather than that of the different form. This was also compatible with the Semi-correspondence type in our revised hierarchy of difficulty. This was the second easy type to interpret. The semi-same characters which are the simplified form of Chinese characters still contain similar features. As mentioned in He (2004), some of the Japanese simplified characters and the Chinese simplified characters are also cognates. Therefore, our Chinese learners might be familiar with these simplified forms and thus perceived them easily. On the other contrary, the participants’ poor performance on the different form could be explained in that our Chinese learners were not familiar with the combinations of different radicals or characters. The different form which is simplified as the fifth Retranslated type is considered relatively difficult compared to the same and the semi-same form. Since this was not the most difficult one in the hierarchy of difficulty, our L2 participants’ performance on the different form was still above the average. As Kess and Miyamoto (1999) suggested, the Japanese kanji characters involve different internal structures and these structures convey different information, such as meanings and pronunciations. Moreover, Japanese people recombine radicals of characters or the whole characters and create new kanji characters and two-kanji characters. Therefore, our participants’ mean scores on the different form were quite low but they could identify the
familiar information from radicals and the reordered kanji characters in two-kanji characters.
In general, our participants’ responses were indeed influenced by the availability of visual information in forms. According to Epp (1969), kanji characters are visual symbols.
When Japanese native speakers learn these characters, they may associate a kanji word as a picture. Chikamatsu (1996) also found that Chinese learners of Japanese applied more visual information when recognizing kanji characters. Hence, our Chinese learners of Japanese behaved similarly to Japanese native speakers in that our participants also counted on visual information.
In addition, the overall scores on the three types of kanji characters for each group are presented in Figure 4-1. All of four groups scored above average on the three types of Japanese kanji characters. The native baseline group performed the best. With regard to our L2 learners, the performances of Group A and Group I were rather satisfactory and the mean scores of three types of form were above 0.7. From the figure, we can also see a regular increase from Group B to Group A.
Figure 4-1 Overall Performances of Each Group on the Three Types of Forms of Japanese Kanji Characters
Besides, Table 4-7 presents the between-group differences among the three types of Japanese kanji characters for each group.
Table 4-7 The p-values for the Three Types of Forms of Japanese Kanji Characters
same form semi-same form different form
F
43.22 17.911 21.919P
.000*** .000*** .000***According to the between-group comparison, Group A (p < .001) and Group I (p < .01) performed significantly better than Group B on the same form. The native baseline group (p
< .001) significantly outperformed all the experimental groups. With regard to the semi-same form, Group A (p < .01) and Group I (p < .05) performed significantly better than Group B.
The native baseline group significantly outperformed Groups B (p < .001), I (p < .01), and A same form semi-same form different form
B M 0.75 0.74 0.54
B SD 0.11 0.12 0.21
I M 0.83 0.82 0.72
I SD 0.07 0.08 0.17
A M 0.88 0.85 0.78
A SD 0.06 0.09 0.14
N M 0.98 0.93 0.91
N SD 0.03 0.06 0.11
0 0.2 0.4 0.6 0.8 1 1.2
(p < .05). As for the different form, Group A (p < .001) and Group I (p < .01) performed significantly better than Group B. The native baseline group performed significantly better than Groups B (p < .001) and I (p < .01).
Hence, the mean scores for each type of form of kanji characters were low for the basic group and the scores became higher for the advanced group, indicating that the interpretation of the three types of forms of Japanese kanji characters varied with the Chinese learners’ L2 proficiency (Fernández Dobao, 2002; Toyoda, 2009). Furthermore, it was also found that Group B performed worse than the other two groups on the three types of forms but their mean scores were still high (i.e., above the average). A possible explanation might be that as Chinese learners of Japanese, they were familiar with the characters to some degree and the visual information was useful for them, even if it was an unfamiliar combination of radicals. Though Group A was different from the native group, their high mean scores for each type indicated that they have acquired the three types of forms. This result matches Toyoda’s (2009) findings that advanced-level learners’
performance was closer to the native group’s performance and that the advanced learners viewed kanji characters as a whole at the character level.
4.2 Meaning-driven Effects
The second research question addresses the issue regarding the influence of the meaning effect (i.e., same vs. different meanings) on our Chinese learners of Japanese. The overall results and general discussion of the findings are presented in the following.
In this section, our revised hierarchy of difficulty is also used to analyze the meaning effect. Meanings are of two types: same and different. The same meaning can be classified as the Correspondence type because the meanings of kanji characters and Chinese characters are corresponding to each other and Chinese learners can transfer meanings into their L2. As for the different meaning, the characters can be found in both languages (in traditional forms and
in simplified forms), but Japanese speakers have different interpretations for these characters.
Thus, the different meaning is classified as the Retranslated type.
Our claim that the same meaning correspondence is at the easiest level and that the different meaning is at the more difficult level can be supported by the results of the present
Our claim that the same meaning correspondence is at the easiest level and that the different meaning is at the more difficult level can be supported by the results of the present